An Integrated Mycobacterial CT Imaging Dataset with Multispecies Information

Area of Science:

• Medical Imaging
• Infectious Diseases
• Artificial Intelligence

Background:

• Global incidence of nontuberculous mycobacterial (NTM) pulmonary disease is rising.
• Current diagnostic methods like bacterial culture are slow, impacting treatment decisions.
• Computed tomography (CT) is a rapid imaging tool for lung lesions, but AI development is limited by small datasets.

Purpose of the Study:

• To address the need for larger datasets in AI-driven NTM diagnosis.
• To create a comprehensive CT dataset for NTM and tuberculosis (TB).
• To facilitate the development of AI algorithms for differential diagnosis of NTM lung disease.

Main Methods:

• Compiled a dataset of 430 NTM and 871 TB computed tomography (CT) cases.
• Included clinical parameters such as demographics, symptoms, and mycobacterial species data.
• Dataset designed to support machine learning applications for improved diagnostic accuracy.

Main Results:

• Established the first comprehensive CT dataset for NTM and TB differential diagnosis.
• Dataset combines imaging data with crucial clinical and etiological information.
• Provides a foundation for training and validating AI models.

Conclusions:

• The developed dataset is crucial for advancing AI-based diagnostic tools for NTM lung disease.
• Enhanced AI algorithms can improve the speed and precision of NTM diagnosis.
• This resource aims to improve patient management and reduce antibiotic misuse in NTM infections.